This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

INA239: ALERT pin assertion time on conversion complete

Part Number: INA239

I wanted to make synchronous ADC measurements of only the shunt voltage with the INA239.
For this i wanted to start the conversion in single-shot mode and use the ALERT pin as an indicator for a finished conversion.

My target timings are 100uS conversion interval with a 84uS conversion time of the ADC.
My SPI speed is at 10MHz.
The ALERT pin is configured in transparent mode, the CNVR bit is set in order to get the ALERT pin asserted when a conversion is finished.

I observed, that the ALERT pin is asserted for around 130uS, which is significantly longer than my desired interval.
This would make it impossible to use the INA239 in my desired configuration.

Can anyone confirm my observed timing on the ALERT pin?
I couldn't find any information about the assertion time of the ALERT pin in the datasheet.

  • Hey Timo,

    Thank you for using the TI forum. 

    I just tested this on the EVM with the same settings you mentioned and saw roughly the same timing as you did. You could try putting it in continuous conversion, then live data will always be available, but I'm guessing that's not what you want.  one other option would be to start the conversion early so that data is available when you want it. (You can even use conversion delay to help with timing). 

    I also tested the alert with continuous conversion, and did get the alert about every 84µs. 

  • Hi Mitch,

    i'm glad i'm not the only one who has observed this.

    I tried the continuous conversion mode and discovered, that after setting the CONFIG_ADC register, the first conversion takes more time than the further continuous ones. But in this mode, the ALERT pin is not asserted for such long time as for the single-shot mode.

    Therefore, i have further investigated the timing.
    Whith every single-conversion time setting, i get a time offset of around 90uS to the set conversion time (observed by polling the CNVRF bit).
    My assumption is, the device goes into power down mode after a single conversion. The device start-up time from shutdown mode is stated in the datasheet with typically 60uS. This would somehow explain my observations.

    My conclusion is, that i have to use the continuous conversion mode, even if there is no synchronisation possible any more, since there is no other setting which will be as fast as i need the values to be updated.

    I also thought about setting the CONFIG_ADC register in sync with my other processing to the continuous conversion mode, but the first conversion will also take much more time until the conversion is complete than the following ones.
    Maybe the chip goes to shutdown mode each time the CONFIG_ADC register is written to?

  • Hello Timo,

    Yes, you are correct, the device does go into shutdown mode after a triggered conversion (see datasheet snip below). Sorry for not catching that sooner, but yes, that does explain the timing observation. 

    From this section it also looks like if you write to the mode bits it will restart a continuous conversion as well, so I believe you could leave it in continuous conversion and write to the MODE with continuous settings to align your data without having to wait for the ADC to wake up. 

  • Hi Mitch,

    i tried your suggestion with leaving the device in continuous mode and writing to the ADC_CONFIG register to sync the conversion, but it will do a restart as well, as stated in your datasheet snip after the highlighted part.
    This means my best option is to leave the device in continuous mode without syncing the conversions.
    Thanks anyways for your support.